comparative value priorities of chinese and new zealand

equivalence and construct validity of the Material Values Scale (MVS, Richins and
Dawson, 1992) is reduced by reversed-scored Likert format items. They conclude that
the cross-cultural applicability of such scales may be enhanced by replacing items posed
as statements with items framed as questions.
Ron Piccolo (2004, personal communication) comments that reverse coded items tend
to be less reliable than their positively phrased counterparts. In many confirmatory
factor analyses of scale items across discipline, reverse coded items tend to exhibit low
scale item reliability. It is not unusual, for example, to read an article where authors had
to drop the reverse coded items in order to get a confirmatory factor model to fit. This
phenomenon could be due to (1) response style of the participant (i.e., participants do
not fully read through the items or do not use the full response scale on reverse coded
items), or (2) the reverse coded items actually measure a characteristic that is not central
to the construct of interest. The art, then, becomes in developing reverse coded items
that are consistent with the construct’s definition. You can imagine a positively phrased
item for leader Consideration, for example that reads, “My leader is considerate of my
feelings” contrasted with a reverse coded item in the same scale, “My leader is not
considerate of my feelings”. The latter phrasing is different from “inconsiderate of my
feelings”, with inconsideration being equated with being actively rude, and not
considerate having an inactive implication. The issue of course is whether being rude is
truly indicative of an absence of consideration or some active process. Thus, the
potential for low reliability of the reverse coded item and subsequently, lower construct
validity.
All that said, reviewers tend to criticise scales that do not utilize reverse coded items. If
you ultimately develop a scale with only positively phrased items, you will certainly
have a scale that achieves desirable psychometric properties, reliability, confirmation in
Standard Error of Measurement, discriminant validity, etc. However, a shortened scale
without reverse coding may not pass the scrutiny of reviewers with more bias than
knowledge. Selmer (2004, personal communication) reports no good experiences with
reverse scored items, indicating they usually distort the factor structure or the reliability
estimates or both. He also suspects that they reduce the response rate in mail surveys
since reverse items make a survey more difficult and time-consuming to complete.
Selmer comments further that if you ask such sensitive questions that you need to check
through complicated reverse-phrased items that the respondents are responding
317